System Specifications

The Customer`s Medical Heart Rate Monitoring System comprises of the following components.

Simulated Link.

The simulated link is built to allow collection of open source bluetooth link logs.

..

Methodology

The business technical goals and trade offs were considered in characterizing the system.The AIC Triad were the most pertinent for the user.It is a venerable method used for analysing end to end network performance.

library(plotly)
library(stats)
library(tidyr)
library(epitools)
library(prettyR)
library(plotrix)
library(ggplot2)
library(reshape2)
library(dplyr)
library(MASS)
library(VIM)
#library(mice)
library(data.table)
library(stringi)
library(stringr)
library(RColorBrewer)
library(DBI)
library(e1071)
library(tidyverse)
library(pastecs)
library(ggpubr)
library(rstatix)
library(car)
library(modeest)
library(DT)
library(expss)
#library(rlist)
library(gdata)
library(lubridate)
library(dbplyr)
library(Hmisc)
library(sqldf)
library(broman)
library(univOutl)
library(ggstatsplot)
library(fitdistrplus)
library(WeibullR)
library(tolerance)
#library(bigmemory)
#library( REdaS)
library(changepoint)
library(inflection)
library(ggpmisc)
library(DescTools)

Results

Input

Returns: Dataframe of the dataset imported from sdl.medicalclaims table Msc database 

#dt_Bt_Link <- read.csv2("bt_Link_kpi_1.csv", header=TRUE, sep=",") 

#setwd("C:/Users/narhuidese/Desktop/USF/HS631_lectures")

dt_Bt_Link <- read.csv2("./data/bt_Link_kpi_1.csv", header=TRUE, sep=",") 
#head(as.data.frame(dt_Bt_Link),n = 10)
datatable(head(dt_Bt_Link,20), rownames = TRUE, options = list(autoWidth = TRUE,pageLength = 7, scrollX=T),caption = 'Table 1: Aggregate link dataset' )
#head(dt_Bt_Link,10)
# datatable(head(dt_Bt_Link,20), rownames = TRUE, options = list(autoWidth = TRUE,pageLength = 7, scrollX=T),caption = 'Table 1: Aggregate link dataset' )
#str(dt_Bt_Link)
#dt_unique <- unique(dt_Emh)
#nrow(dt_unique)
#rm(dt_unique)
dt_Bt_Link_1 <- dt_Bt_Link

Restructure the datatypes

dt_Bt_Link_1$BT_MAC_Throughput_DL.1 <- as.numeric(as.character(dt_Bt_Link_1$BT_MAC_Throughput_DL.1))
dt_Bt_Link_1$Tx_Power_.dBm._class.2  <- as.numeric(as.character(dt_Bt_Link_1$Tx_Power_.dBm._class.2 )) #as.numeric(as.character(x))
dt_Bt_Link_1$Rx.Power..dBm. <- as.numeric(as.character(dt_Bt_Link_1$Rx.Power..dBm.))
dt_Bt_Link_1$Physical_Layer.PHY.  <- as.character(dt_Bt_Link_1$Physical_Layer.PHY. )
dt_Bt_Link_1$Link_Margin <- as.numeric(as.character(dt_Bt_Link_1$Link_Margin))
dt_Bt_Link_1$Signal_to_Noise_Ratio  <- as.numeric(as.character(dt_Bt_Link_1$Signal_to_Noise_Ratio ))
dt_Bt_Link_1$Max_number_of_retransmission  <- as.factor(dt_Bt_Link_1$Max_number_of_retransmission)

Returns: Change structure of the data variables

#str(dt_Bt_Link_1)
#str(dt_Emh1)
#head(dt_Bt_Link_1,50)

Check for NA`S

 any(is.na(dt_Bt_Link_1))
## [1] FALSE

Availability

Availability is dependent on uptime & downtime that may be due to any to possible equipment outage,power failure or fading.The link was designed to be a high reliability link of “five nines”.

Availability Metric: Mean time between failure. MTBF/(MTBF+MTTR) * The receiver must have a sensitivity level for which the bit error rate (BER) 0.1% is met.

Distribution of Rx Lev at the input of the receiver detector. Create group variable for link performance. Categorize RXlev by noise floor

#if RxLev > Sensitivity + link margin threshold = Uptime
#if link margin threshold > Rxlev & Rxlev > Sensitivity = danger zone
#if  RxLev < sensitivity = unintelligible transmission.

#Categorize the  RXlev by noise floor into 2 groups
dt_Bt_Link_1$RxLev_cat[dt_Bt_Link_1$Rx.Power..dBm. > (dt_Bt_Link_1$Rx_Sensitivity_.dBm)] = "Uptime"
dt_Bt_Link_1$RxLev_cat[dt_Bt_Link_1$Rx.Power..dBm. <= (dt_Bt_Link_1$Rx_Sensitivity_.dBm.)] = "Downtime"
#Categorize the Sensitivity into two groups
dt_Bt_Link_1$Sensitivity_cat[dt_Bt_Link_1$Rx_Sensitivity_.dBm. > -90] = "Sensitivity_drop"
dt_Bt_Link_1$Sensitivity_cat[dt_Bt_Link_1$Rx_Sensitivity_.dBm. <= -90]= "Sensitivity_normal"
#head(dt_Bt_Link_1)
table(dt_Bt_Link_1$RxLev_cat)
## 
## Downtime   Uptime 
##        9     3727

Pie chart of performance of the Link

Returns: Frequency table of link performance

sort(table(dt_Bt_Link_1$RxLev_cat))
## 
## Downtime   Uptime 
##        9     3727
Av <- as.data.frame(table(dt_Bt_Link_1$RxLev_cat))

___Returns: Plot of distribution of availability.

p<- ggplot(Av, aes(x="Link Availability", y=Freq, fill=Var1))+
geom_bar(width = 1, stat = "identity")
pie1 <- p + coord_polar("y", start=0)
pie1
Fig 1:Frequency distribution of the denominator physicians

Fig 1:Frequency distribution of the denominator physicians 

#Availabilty = Uptime/(uptime + downtime) 
Availability = (Av$Freq[2]/(Av$Freq[2]+Av$Freq[1]))*100
Availability
## [1] 99.7591

Returns: .Number of times that Rxlev was at the certain value._ The confidence interval shown for the system parameter bellow assumes normal distribution. 95% confidence interval of the RxLev is the range that contains the true value of the estimated parameter.

# for reproducibility
set.seed(123)

# plot
histmr <- gghistostats(
  data = dt_Bt_Link_1,
  x = Rx.Power..dBm.,
   binwidth = 0.6,
  #y=stat(count)/sum(stat(count)),
  normal.curve = TRUE, # superimpose a normal distribution curve
  normal.curve.args = list(color = "red", size = 1),
  conf.level = 0.95,
  bar.measure = "proportion",
  k=2,
  #type ="nonparametric",
  type = "robust",
  robust.estimator ="mom",
  #bar.fill = "green",
  bin.args = list(color = "black", fill = "#0088B1FF", alpha = 0.7),
  #ggtheme = ggstatsplot::theme_ggstatsplot(),
  ggtheme = ggplot2::theme_gray(),
  ggplot.component = list(theme(axis.text = element_text(colour = "blue"),text = element_text(size=15),plot.title = element_text(color = "red", size = 14, face = "bold"),plot.subtitle = element_text(color = "blue",size = 12)),
  ylab("Percentage of time"), xlab("Rx Power Level "),ggtitle(labs(title = "Distribution of RxLev"))),
  messages = FALSE
)
histmr
Fig 9:Histogram of measure rate with summary stats

Fig 9:Histogram of measure rate with summary stats 

# p + theme(
#   plot.title = element_text(color = "red", size = 12, face = "bold"),
#   plot.subtitle = element_text(color = "blue"),
#   plot.caption = element_text(color = "green", face = "italic")
# )
#dt_IMgn1$MarketScan_Specialty <- vlookup(dt_IMgn1$ProviderNPI,dt_phy_spec,result_column = 2,lookup_column = 1)
#head(dt_Aih1)
#dt_Emh1$MarketScan_Specialty <- vlookup(dt_Emh1$ProviderNPI,dt_phy_spec,result_column = 2,lookup_column = 1)
#head(dt_Emh1)
#str(dt_IMgn1)

Returns: Probability Density of RxLev grouped by avaialbility Which rxLev value contribute the most to avaialability?

#age by mortality
Densityplot <- ggplot(data = dt_Bt_Link_1,aes(x=Rx.Power..dBm.,fill=RxLev_cat))+
  geom_density(alpha = 0.7,cex=1)+ggtitle(labs(title = "Distribution of RxLev grouped by Availability"))+theme(text = element_text(size=10))
Densityplot

Returns: Distribution of Receive level by avaialbility

#INR value grouped by icd9 categories
Histogram2 <- ggplot(data = dt_Bt_Link_1,aes(x=Rx.Power..dBm.,fill=RxLev_cat))+
  geom_histogram(alpha = 0.7,binwidth = .8,cex = 6,position = "dodge")+ggtitle(labs(title = "Histogram of Rxlev grouped by availability"))+theme(text = element_text(size=10))
Histogram2

Returns: Plot of sensitivity by availability

#INR value grouped by icd9 categories
Histogram3 <- ggplot(data = dt_Bt_Link_1,aes(x=Rx_Sensitivity_.dBm.,fill=RxLev_cat))+
  geom_histogram(alpha = 0.7,binwidth = .1,cex = 3,position = "dodge")+ggtitle(labs(title = "Histogram of Sensitivity grouped by Availability  "))+theme(text = element_text(size=10))
Histogram3

Histogram2

Histogram3

First summarise and transform your data:gender GROUPED BY rxlev cat ___Returns: A plot of the impact of sensitivity on the outage or downtime.

d4 <- dt_Bt_Link_1 %>% 
  group_by(Sensitivity_cat,RxLev_cat) %>% 
  summarise(count=n()) %>% 
  mutate(perc=count/sum(count))
## `summarise()` has grouped output by 'Sensitivity_cat'. You can override using the `.groups` argument.
bar4 <- ggplot(data=d4, aes(x=Sensitivity_cat, y=perc, fill=RxLev_cat)) + 
  geom_bar(position = 'dodge', stat ='identity')+scale_y_continuous(labels = scales::percent)+labs(x = "Sensitivity", y = "Percentage of availability", fill = "Availability") + ggtitle(labs(title = "Device Sensitivity by availability"))+ theme(text = element_text(size=10))+ geom_text(aes(label = scales::percent(perc),y = perc),position=position_dodge(width=0.7), vjust = -0.25,size=3.5)
bar4  + facet_grid(. ~ Sensitivity_cat,scales='free')

Integrity

Throughput: Rx Device gauranteed throughput based on QOS - quality of service of the system. This system has a capacity of 2Mbps. As seen bellow,the through put is not consistent.

Returns: A plot of throughput delivered with respect to time

# Multiple line plot
ggplot(dt_Bt_Link_1, aes(x = Time_.sec., y = BT_MAC_Throughput_DL.1)) + 
  geom_line(aes(color = RxLev_cat), size = 1.5)  + ggtitle(labs(title = "Bluetooth Device downlink throughput"))+ theme(text = element_text(size=10))

Confidentiality

Returns: Number of Descriptive Statistics of Receive Power.

Assumption: Normal distribtution because of the Central Limit theorem for large dataset where n >> 30

set.seed(1243)
x<-as.data.frame(stat.desc(dt_Bt_Link_1$Rx.Power..dBm., basic=TRUE, desc=TRUE, norm=TRUE, p=0.95))

x$`stat.desc(dt_Bt_Link_1$Rx.Power..dBm., basic = TRUE, desc = TRUE, norm = TRUE, p = 0.95)` <- round(x$`stat.desc(dt_Bt_Link_1$Rx.Power..dBm., basic = TRUE, desc = TRUE, norm = TRUE, p = 0.95)`, 2)

names(x)[1] <- "Value"
names(x)[0] <- "Sample Stats"
x
Value
nbr.val 3736.00
nbr.null 0.00
nbr.na 0.00
min -88.08
max -41.95
range 46.13
sum -286314.34
median -80.56
mean -76.64
SE.mean 0.16
CI.mean.0.95 0.31
var 90.64
std.dev 9.52
coef.var -0.12
skewness 1.63
skew.2SE 20.39
kurtosis 1.90
kurt.2SE 11.86
normtest.W 0.76
normtest.p 0.00

Returns: Number of Physicians with denominator cases and numerator cases.

The BLE encryption protocol doesn`t is not shown in the wireshrk Bluettoth protocols. I have not seen it yet. Currently, the encryption scheme used for this link has not been identified.

Conclusion

  • The results show that capacity utilization is less than 50%.
  • Average RxLev is -70dBm.
  • The receiver experiences desensitization possibly due to hardware failure.
  • Currently the system availability can be gauranteed at 99.759% of the time.
 

For project use Project Presentation

Project